The reactions OASST2 records on the messages in oasst_messages, in long form: 2,469 rows over 1,294 of the messages, one per message and name, with the number of people who used it.
Format
A tibble with 2,469 rows and 3 columns:
- message_id
The message, as in oasst_messages.
- reaction
The name OASST2 records.
- n
How many times it was used on the message, an integer.
Source
As oasst_messages: the emojis field of the same OASST2 file,
Apache License 2.0, processed by data-raw/oasst.R.
Details
The names are OASST2's own, and they are realistically messy. "+1"
(2,323 uses), "-1" (802) and "red_flag" (48) are reactions, the first
two under GitHub's names. "_skip_reply" (1,365), "_skip_labeling" (857)
and "_skip_ranking" (396) are not: they are the interface's skip actions,
stored beside the reactions, and kept here because a real tally holds such
things and an analysis has to decide what to do with them.
emoji_from_counts() refuses a name it cannot resolve, by name, rather
than counting it as nothing.
See also
emoji_from_counts(), which turns a tally into text the other
verbs read.
Examples
head(oasst_reactions)
#> # A tibble: 6 × 3
#> message_id reaction n
#> <chr> <chr> <int>
#> 1 c5d1ac81-3370-44f6-bed6-ed02a016bb72 -1 1
#> 2 d121645a-6d6e-488e-bc4a-6208f93109c7 +1 1
#> 3 d121645a-6d6e-488e-bc4a-6208f93109c7 -1 1
#> 4 e7aa0bcf-82af-4589-b903-4af455b058bf -1 1
#> 5 960f1b99-9433-41c0-abf1-a2331a4773ed +1 1
#> 6 960f1b99-9433-41c0-abf1-a2331a4773ed -1 9
# the reactions alone, as text every verb can read
votes <- oasst_reactions[oasst_reactions$reaction %in% c("+1", "-1"), ]
emoji_frequency(emoji_from_counts(votes, reaction, n, id = message_id), text)
#> # A tibble: 2 × 5
#> emoji name shortcode group n
#> <chr> <chr> <chr> <chr> <int>
#> 1 👍 thumbs up +1 People & Body 2323
#> 2 👎 thumbs down -1 People & Body 802